Automated surface analysis systems for detecting surface flaws and particle contamination have many applications, particularly in semiconductor manufacture and research. For example, surface analysis systems can be used to detect surface flaws and contaminating particles on silicon wafers; such flaws and contamination tends to interfere with processing and to reduce the quality of the completed semiconductor device.
Current automated surface analysis systems detect surface defects by light that has been scattered or deflected from an intense illuminating beam directed at the surface being analyzed. The illuminating beam is commonly a focused spot of laser light which is scanned across the surface being inspected. The WIS-600 surface analysis system manufactured by Estek Corporation is one example of current surface analysis systems.
Scanned laser surface analysis systems include a scattered light detector assembly for detecting laser light scattered by specific defects. Scattered-light detectors are particularly useful in detecting surface particles since such particles scatter light. Surface analysis systems may also include an assembly which directly detects reflected laser light for detecting surface defects that cause deviations in surface flatness. The two types of detectors, used together, may detect a broad range of surface defects and contamination.
Referring now to surface particle detection by a surface analysis scanner, light scattered by particles on the scanned surface is collected and converted to electrical signals which are conditioned and then processed by a computer to determine the desired surface information. Threshold signal level and device gain determine the sensitivity of the device to optical signals received by the scattered light detector. The threshold signal level refers to a minimum electrical signal required to produce a response on the system's output device. The lower the threshold setting, the higher the sensitivity of the system to detected light. The gain of the surface analysis device refers to the relationship between the amplitude of the electrical signal generated by the system and the optical signal amplitude received by the detector which converted the optical signal into the electrical signal. The higher the system gain, the greater the signal produced for a given optical signal.
Although the relationship between light scattering amplitude and particle size is not linear or single valued, it is clear that increasing the gain of an instrument changes the size of particle that can be detected by the instrument. Thus, it is critical to meaningful particle counting to hold the system gain constant.
A reference scattering target or calibration target is needed to adjust the instrument gain to a constant level. Such a target must produce reliably repeatable signals and thus must not produce unintentional signals from particle contamination of the target surface.
Once the gain relative to the threshold settings is fixed, what remains is to ensure that all legitimate signals occurring above the desired thresholds, and only those signals, are counted and that some meaningful particle attribute is ascribed to the threshold settings. Again, reference targets having known artifacts or reference features are indispensible in ensuring that only what should be counted is counted. Clearly, contamination of such a reference target by surface particles could disturb the count of reference features.
Several ambiguities hamper the interpretation of light scatter data. The first ambiguity arises from the fact that the relationship between particle size and the amplitude of scattered light is neither linear nor single-valued. Furthermore, there is no existing exact mathematical solution to the problem of relating the portion of incident light scattered by a particle resting on a surface to the particle size or shape. Also it is well documented that the optical properties of a substrate have a significant effect on the optical sizing of a particle. Due to these ambiguities, characterizing parameters are necessary in order meaningfully to compare data from surface analysis instruments.
With regard to scattered light particle detectors, there are seven characterizing parameters of interest. These parameters are sensitivity, counting accuracy, uniformity, dynamic range, spatial resolution, repeatability, and stability. Thus, in addition to fixing system gain relative to the threshold settings, each of the characterization parameters must be determined for a particular instrument in order for its data to be meaningful.
As with gain and threshold setting or calibration, the determination of characterization parameters requires calibration targets having reference features with known light scattering properties. Several types of calibration targets are in current use, including wafers having reference particles dispersed on their surface or a portion thereof, or wafers with point features etched into their surface to simulate effectively a patterned array of particles.
Regardless of the type of calibration target employed, it is crucial that the calibration target itself be free of unintentional particle contamination. If particles other than the reference features exist on the calibration target, light scattered by these unknown particles will be detected along with the light scattered from the known features and will make calibration or parameter determination less precise and less accurate. Thus, a great deal of care must be taken to ensure that current calibration targets are free from contamination which might upset the setting of system gain and thresholds, as well as the determination of the instrument's characterizing parameters.